Audit Quality, Corporate Governance, and Earnings Management: A Meta-Analysis ija_403 57 78 Jerry W. Lin 1 and Mark I. Hwang 2 1 University of Minnesota Duluth 2 Central Michigan University Earnings management is of great concern to corporate stakeholders. While numerous studies have investigated the effects of various corporate governance and audit quality variables on earnings management, empirical evidence is rather inconsistent. This meta-analysis identifies 12 significant relationships by integrating results from 48 prior studies. For corporate governance, the independence of the board of directors and its expertise have a negative relationship with earnings management. Similar negative relationships exist between earnings management and the audit committee’s independence, its size, expertise, and the number of meetings. The audit committee’s share ownership has a positive effect on earnings management. For audit quality, auditor tenure, auditor size, and specialization have a negative relationship with earnings management. Auditor independence, as measured by fee ratio and total fee, is also a deterrent to earnings management. Key words: Audit committee, audit quality, auditor choice, corporate governance, earnings management, fraud, independence, meta-analysis SUMMARY Earnings management is of great concern to corporate stakeholders. While numerous studies have investigated the effects of various corporate governance and audit quality variables on earnings management, empirical evidence of their effects is rather inconsistent. This paper applied meta-analytic techniques to empirical data from 48 studies that examined relationships between corporate governance and audit quality variables and earnings management. Of the 17 relationships tested, 12 showed significant effects. Specifically, for corporate governance, the independence of the board of directors and its expertise have a negative relationship with earnings management. Similar negative relationships exist between earnings management and the audit committee’s independence, its size, expertise, and the number of meetings. The audit committee’s share ownership is positively related to earnings management. For audit quality, auditor tenure, auditor size, and auditor specialization have a Correspondence to: Jerry W. Lin, Department of Accounting, University of Minnesota Duluth, 1318 Kirby Drive, LSBE-360F, Duluth, MN 55812, USA. Email: jlin@d.umn.edu International Journal of Auditing doi:10.1111/j.1099-1123.2009.00403.x Int. J. Audit. 14: 57–77 (2010) ISSN 1090-6738 © 2009 Blackwell Publishing Ltd, 9600 Garsington Rd, Oxford OX4 2DQ, UK and Main St., Malden, MA 01248, USA. negative relationship with earnings management. Auditor independence, as measured by fee ratio and total fee, is also a deterrent to earnings management. Relationships that were non-significant include the effects of the board of directors’ stock ownership, existence of an audit committee, and the separation of the board chairperson position from the CEO position. These are potential areas for future research. Another prospective research stream is to examine the moderating effect of certain variables such as country. For example, we found that while the independence of the board of directors is significant overall, the effect is more profound in countries other than the United States. The opposite is observed about the independence of the audit committee: the effect is more pronounced in the United States than in other countries. Similarly, the effect of auditor size as a deterrent to earnings management is more significant in the United States than in other countries. In fact, the effect is not significant when data from other countries are tested. On the other hand, the share ownership of the board of directors has no significant effect on earnings management, in either the United States or other countries. We also tested the effects of two levels of audit committee independence (complete and proportional independence), and found both to have a significant effect in reducing earnings management. Similarly, future research can examine the moderating effect of important regulations, such as the Sarbanes-Oxley Act, by comparing data from pre- and post-SOX. INTRODUCTION Much research has been conducted on the determinants of earnings management such as a firm’s financial characteristics, corporate governance and audit quality. However, the extant studies have reported mixed results. The purpose of this paper is to use meta-analysis techniques to synthesize and evaluate the findings from the large number of existing studies on the determinants of earnings management. Our focus is on the effect of corporate governance effectiveness and audit quality. Meta-analysis is the application of statistical methods to a large collection of results from existing individual studies for the purpose of integrating and evaluating the research findings. Use of meta-analysis often makes it possible to reach stronger conclusions or more valid inferences about a common research issue than in a narrative literary review (Wolf, 1986). The remainder of this paper is organized as follows. The next section provides an overview of prior research on the relationships between earnings management and corporate governance and audit quality. Next, we describe the research methodology, followed by discussions of meta-analytic results. The last section presents concluding remarks. LITERATURE REVIEW Earnings management Various definitions exist for earnings management. Schipper (1989) appears to have captured the essence of earnings management by defining it as ‘purposeful intervention in the external financial reporting process with the intent of obtaining private gain’. Likewise, Healy & Wahlen (1999) state that ‘earnings management occurs when managers use judgment in financial reporting and in structuring transactions to alter financial reports to either mislead some stakeholders about the underlying economic performance of the company or to influence contractual outcomes that depend on reported accounting numbers.’ Regardless of the definition adopted, earnings management is inherently unobservable. Most prior studies use various measures of discretionary or abnormal accruals as proxies for earnings management. Other measures used include earnings restatement and financial reporting fraud. A regression model is typically employed to investigate the effects of various independent variables on earnings management in the form of: EM X X X kN kkkiikk =+ + ++ + = ββ β β ε 011 22 12 ,, , , , , , (1) where EM is earnings management and X i represents either a control variable or an independent variable under investigation in study k in a set of N prior studies in a meta-analysis. Next, we provide an overview of research on the effects on earnings management of factors relating to corporate governance effectiveness and quality of external audit. Corporate governance Owing to the separation of ownership and control (and the resulting agency problems) in the modern business world, a system of corporate governance 58 J. W. Lin and M. I. Hwang Int. J. Audit. 14: 57–77 (2010)© 2009 Blackwell Publishing Ltd is necessary, through which management is overseen and supervised to reduce the agency costs and align the interests of management with those of the investors. While there is no generally accepted definition, corporate governance may be defined as a system ‘consisting of all the people, processes and activities to help ensure stewardship over an entity’s assets’ (Messier et al., 2008: 36). A good corporate governance structure helps ensure that the management properly utilizes the enterprise’s resources in the best interest of absentee owners, and fairly reports the financial condition and operating performance of the enterprise. For corporations in the US, the body primarily responsible for management oversight is the board of directors and its designated committees. The audit committee, consisting of members of the board, assists the board in its oversight of the financial reporting process. The role of the corporate governance structure in financial reporting is to ensure compliance with generally accepted accounting principles (GAAP) and to maintain the credibility of corporate financial statements. The corporate governance mechanisms that are the focus of recent regulations and prior studies are attributes related to the organization and functioning of the board in general and its audit committee in particular. Properly structured corporate governance mechanisms are expected to reduce earnings management because they provide effective monitoring of management in the financial reporting process. Unfortunately, empirical research to date provides inconsistent evidence on the relationship between measures of corporate governance effectiveness and earnings management (earnings quality or the lack thereof). For example, while Davidson et al. (2005) and Klein (2002) report a significantly negative relationship between board independence and earnings management, Park & Shin (2004) and Peasnell et al. (2005) fail to find any significant relationship. Such inconsistency also exists in empirical evidence on the relationships between earnings management and other attributes related to board effectiveness in monitoring management in the financial reporting process. Often the board of directors delegates work on important tasks to its standing committees. For example, the audit committee is charged with overseeing financial reporting. The audit committee’s primary role is to help ensure high quality financial reporting by the firm. Therefore, a properly structured and functioning audit committee is expected to reduce opportunistic earnings management. A number of recent studies examine the effect of an audit committee’s characteristics on earnings management but have provided mixed evidence as is the case in research on effectiveness of the board of directors in reducing earnings management. For example, while Abbott et al. (2000) document that occurrence of earnings management decreases with independence of the audit committee, Choi et al. (2004) find no such effect. Also, Xie et al. (2003) find no significant association between the number of directors on the audit committee and earnings management. Similarly, Abbott et al. (2004) find no impact of audit committee size on earnings restatements. In contrast, Yang & Krishnan (2005) report that audit committee size is negatively associated with earnings management (using abnormal accrual as proxy), implying that a certain minimum number of audit committee members may be relevant to quality of financial reporting. There is also concern that compensating audit committee directors with stock and stock options may result in impairment of their independence (Millstein, 2002); however, empirical evidence on this issue has been limited until recently. Bédard et al. (2004) document that the more stock options that can be exercised in the short run relative to the total of options and stocks held by audit committee directors, the higher the likelihood of aggressive earnings management. Yang & Krishnan (2005) report that stock ownership by board members on the audit committee is positively associated with earnings management. These results contradict the findings by Beasley (1996) that the likelihood of fraud decreases as stock ownership by outside directors (not necessarily audit committee directors) on the board increases. Audit quality The agency problems associated with the separation of ownership and control, along with information asymmetry between management and absentee owners, create the demand for external audit. External auditors are responsible for verifying that the financial statements are fairly stated in conformity with GAAP and that these statements reflect the ‘true’ economic condition and operating results of the entity. Thus, the external auditor’s verification adds credibility to Audit Quality, Corporate Governance, and Earnings Management: A Meta-Analysis 59 Int. J. Audit. 14: 57–77 (2010)© 2009 Blackwell Publishing Ltd the company’s financial statements. Also, the external auditors are required by auditing standards to discuss and communicate with the audit committee about the quality, not just the acceptability, of accounting principles applied by the client company. Therefore, a quality audit is expected to constrain opportunistic earnings management as well as to reduce information risk that the financial reports contain material misstatements or omissions. The guidelines and measures for the quality of the external auditor’s performance are set forth in generally accepted auditing standards, such as competence, independence and exercise of due professional care. Obviously, the quality of the auditor’s performance is multi-dimensional as set forth in the auditing standards, and differences in audit quality are to be expected. ‘Audit quality differences result in variation in credibility offered by the auditors, and in the earnings quality of their audit clients. Because auditor quality is multidimensional and inherently unobservable, no single auditor characteristic can be used to proxy for it’ (Balsam et al., 2003: 71). Since audit quality may be affected by a number of factors, it is not surprising that researchers have used various measures to proxy for audit quality in prior studies. For example, researchers have examined the effects of auditor brand name (auditor size) and industry specialization, auditor tenure, provision of various services by the auditor and auditor independence on a number of issues directly or indirectly related to financial reporting. Empirical evidence on these audit quality measures has been mixed. For example, while many existing studies show that the use of brand name (i.e., Big 4/5/6) auditors reduces earnings management (e.g., Becker et al., 1998; Francis et al., 1999; Lin et al., 2006), many others fail to report such findings (e.g., Bédard et al., 2004; Davidson et al., 2005). As another example, Frankel et al. (2002) report that the ratio of non-audit service fees to total auditors’ fees (proxy for impaired auditor independence) is positively associated with small earnings surprises and with the magnitude of discretionary accruals (proxies for earnings quality or earnings management). Their results provide support to the SEC’s position that non-audit fees can impair auditor independence and hence audit quality. On the other hand, Chung & Kallapur (2003) find no significant relationship between discretionary accruals and audit fees or non-audit fees. Similarly, Raghunandan et al. (2003) find no evidence supporting the claim that non-audit fees or total fees inappropriately influence the audit of financial statements that are subsequently restated. Inconsistent results reported in prior studies about the effects of the other factors affecting audit quality on earnings quality are highlighted in the results section below. METHODOLOGY The first step in a meta-analysis is to locate relevant studies through computer and manual searches. Various combinations of key words are used to search commonly available computerized literature databases, such as ABI/Inform and Business Source Premier, to locate empirical (archival) studies that deal with earnings management. Key words used include earnings, accrual, restatement, fraud, management, quality, audit and governance. The computer searches were conducted from late October to early November 2007 and publications available online or in print were then reviewed for possible inclusion. References in studies identified in computer searches were also scanned to find additional studies. Published articles had to be empirical archival studies that met all of the following criteria for inclusion in the meta-analysis: (1) the dependent variable must be measures based on abnormal (discretionary) accrual, financial statement restatement or reporting fraud; (2) the independent variables in the empirical multivariate model must include measures of audit quality, or effectiveness of corporate governance relating to the board of directors or audit committee; and (3) the test statistics or p-value needed to compute the effect size must be reported. Ultimately, 48 studies were included in the meta-analysis. Table 1 lists these studies and also provides information about the dependent and independent variables, and data years and countries for sample firms used. Many of these included studies measure the same variable in multiple ways. For example, audit committee independence can be measured by its membership that is made of 100 percent outsiders (i.e., completely independent) or over 50 percent outsiders (i.e., proportionally independent) (Klein, 2002). Multiple results from the same study are combined to satisfy the independent sample requirement for meta-analysis. We also examine sample size, data years and countries used in the included studies to ensure no later studies use identical data from any of the prior publications. A 60 J. W. Lin and M. I. Hwang Int. J. Audit. 14: 57–77 (2010)© 2009 Blackwell Publishing Ltd Table 1: Studies included Study Independent variable Dependent variable Data country/year Abbott et al. 2000 audit committee characteristics fraud US 1980–96 Abbott et al. 2004 audit committee characteristics restatement US 1991–99 Agrawal & Chadha 2005 auditor fees; board & audit committee characteristics restatement US 2000–01 Antle et al. 2006 auditor fees abnormal accrual UK 1994–2000 Ashbaugh et al. 2003 non-audit service abnormal accrual US 2001 Balsam et al. 2003 auditor specialization abnormal accrual US 1991–99 Bauwhede et al. 2003 auditor size discretional accrual Belgium 1991–97 Bauwhede & Willekens 2004 auditor size discretional accrual Belgium 1994–96 Beasley 1996 audit committee characteristics fraud US 1979–90 Becker et al. 1998 audit committee and BOD characteristics abnormal accrual US 1993 Bédard et al. 2004 audit committee characteristics abnormal accrual US 1996 Benkel et al. 2006 BOD & audit committee characteristics discretional accrual Australia 2001–03 Carey & Simnett 2006 audit partner tenure abnormal accrual Australia 1995 Chen et al. 2005 auditor size & specialist discretional accrual Taiwan 1999–2002 Choi et al. 2004 audit committee characteristics abnormal accrual Korea 2000–01 Chung et al. 2005 audit committee & BOD characteristics abnormal accrual US 1980–96 Cormier & Martinez 2006 BOD & auditor size discretional accrual France 2000–02 Crutchley et al. 2007 BOD characteristics fraud US 1991–2002 Davidson et al. 2005 audit committee characteristics abnormal accrual Australia 2000 Ferguson et al. 2004 non-audit service abnormal accrual UK 1996–98 Firth et al. 2007 board characteristics & auditor size discretional accrual China 1998–2003 Francis et al. 1999 auditor size discretionary accrual US 1975–94 Frankel et al. 2002 non-audit fee abnormal accrual US 2001 Gul et al. 2002 audit committee characteristics abnormal accrual Australia 1992–93 Hoitash et al. 2007 auditor fees & size discretional accrual US 2000–03 Huang et al. 2007 auditor fees & size discretional accrual US 2003–04 Jaggi & Leung 2007 BOD characteristics & auditor size discretional accrual Hong Kong 1999–2000 Jaggi & Tsui 2007 BOD characteristics discretional accrual Hong Kong 1995–99 Jeong & Rho 2004 auditor size discretional accrual Korea 1994–98 Kao & Chen 2004 BOD size & characteristics discretional accrual Taiwan; year not reported Klein 2002 audit committee and BOD characteristics abnormal accrual US 1992–93 Krishnan 2003a auditor specialization abnormal accrual US 1989–98 Lee et al. 2006 corp governance discretional accrual US 1991–2004 Li & Lin 2005 non-audit fee restatement US 2000 Lin et al. 2006 audit committee characteristics restatement US 2000 Maijoor & Vanstraelen 2006 auditor size discretional accrual France, Germany, UK 1992–2000 Menon & Williams 2004 audit committee characteristics abnormal accrual US 1998–99 Mitra 2007 auditor fees discretional accrual US 2000 Myers et al. 2003 auditor tenure abnormal accrual US 1988–2000 Park & Shin 2004 board characteristics abnormal accrual Canada 1996–97 Peasnell et al. 2005 audit committee & board characteristics abnormal accrual UK 1993–96 Piot & Janin 2007 audit committee & auditor characteristics discretional accrual France 1999–2001 Raghunandan et al. 2003 non-audit fee restatement US 2001 Rahman & Ali 2006 BOD & audit committee char and auditor size discretional accrual Malaysia 2002–03 Reitenga & Tearney 2003 BOD & audit committee characteristics discretional accrual US 1987–96 Reynolds et al. 2004 auditor fees & size discretional accrual US 2000 Van der Zahn & Tower 2004 audit committee characteristics abnormal accrual Singapore 2000–01 Yang & Krishnan 2005 audit committee characteristics abnormal accrual US 1996–2000 Note: Studies included must meet all of the following criteria: 1. The dependent variable must be measures based on abnormal (discretionary) accrual, financial statement restatement or reporting fraud. 2. The independent variables in a multivariate model must include measures of audit quality and effectiveness of corporate governance relating to the board of directors and audit committee. 3. The test statistics or p-value for computing the effect size must be reported. Audit Quality, Corporate Governance, and Earnings Management: A Meta-Analysis 61 Int. J. Audit. 14: 57–77 (2010)© 2009 Blackwell Publishing Ltd large number of studies are excluded from the meta-analysis because they do not meet the criteria specified above. Table 2 lists these excluded studies and states the reason for their exclusion. Following prior meta-analysis studies in accounting (e.g., Hay et al., 2006; Kinney & Martin, 1994), we use the Stouffer combined test to summarize the effects on earnings management of various independent variables, which are reported with a t-statistic, c 2 -statistic, or p-value in individual existing studies. We convert all t-statistics and c 2 -statistics to their corresponding p-values and then to Z-statistics as the measure of effect size. The individual Z-statistics are then combined using the following formula (Wolf, 1986: 20): Unweighted Zc Z N = ∑ , (2) where N is the number of studies under review. It may be argued that not all studies in a meta-analysis should be given equal weight. Some studies may use a small sample, while others may be based on a much larger sample. In the unweighted case, as is the case in Formula (2) above, studies with small samples could exert a much stronger effect on the results than warranted. Wolf (1986) recommends that both the unweighted and weighted Zc be calculated. Therefore, the Stouffer combined test based on the sample-size weighted Zc giving more weight to large samples is calculated as follows (Wolf, 1986: 40): Weighted Zc df Z df = ∗ ∑ ∑ 2 , (3) where df is the degrees of freedom associated with the statistic of each study. Finally, there is a potential problem when including only published studies. While the manuscript review process helps ensure the quality of published studies, a publication bias may result from the tendency that studies with significant results or larger effect sizes are more likely to be published than those without significant results or with smaller effect sizes. This problem is commonly referred to as the ‘file drawer’ problem in that these unpublished studies are buried away in ‘file drawers’ (Hay et al., 2006; Wolf, 1986). To deal with publication bias, in a meta-analysis, the fail-safe number, N fs , is calculated to show the number of studies failing to report significant results that would be needed to reverse a conclusion about a significant relationship between the dependent and independent variables. Using the results of the Stouffer combined test, the fail-safe number is computed as follows (Rosenthal, 1991: 261): N kkz fs = ××− ⎛ ⎝ ⎜ ⎞ ⎠ ⎟ (. . , 2 2 706 2 706 (4) where k is the number of studies in themeta-analysis and z is the combined standard z-value for the meta- analysis. The robustness of a significant relationship as represented by its fail-safe number can be compared with a critical number of studies that may be filed away. Rosenthal (1991: 262) provides the following equation for calculating the critical number of studies: Critical number of studies 5 10,=× ( ) +k (5) where k is the number of studies in the meta-analysis. According to Clark-Carter (1997) and Rosenthal (1991), the file drawer issue is only a problem if the fail-safe number is not greater than the critical number of studies. Moreover, the fail-safe number and the critical number of studies only need be calculated for significant relationships. RESULTS Table 3 reports the main results of the meta-analysis of the effects of corporate governance and audit quality attributes on earnings management. For each attribute, we discuss its nature and hypothesized effect on earnings management (earnings quality or lack thereof), and the results from our meta-analysis. Corporate governance The role of corporate governance structure of a corporation in financial reporting is to ensure compliance with GAAP and to maintain the credibility of corporate financial statements. Common measures of corporate governance effectiveness that are the focus of prior research are related to the composition, expertise, and activity of the board of directors (BOD) and its audit committee (AC). BOD independence Fama & Jensen (1983) recognize the board of directors as the most important management 62 J. W. Lin and M. I. Hwang Int. J. Audit. 14: 57–77 (2010)© 2009 Blackwell Publishing Ltd Table 2: Studies excluded Study Different independent variable Different dependent variable No applicable data Abbott et al. 2006 discretional accruals audit fees Aboody et al. 2005 insider trading Adjaoud et al. 2007 accounting rate of return & market-based performance Aier et al. 2005 CFO characteristics Akhigbe et al. 2005 abnormal stock return Arnold et al. 2001 additional audit work Arthaud-Day et al. 2006 turnover of CEO, CFO, BOD and audit committee members Ascioglu et al. 2005 market liquidity Ashbaugh-Skaife et al. 2006 credit ratings Ball & Shivakumar 2005 firm characteristics Ball & Shivakumar 2006 cash flows & stock returns Bartov et al. 2001 audit opinions Beasley et al.2000 simple t-tests Beatty & Weber 2006 goodwill writeoff Beekes et al. 2004 reporting conservatism Blouin et al. 2007 selection of new auditors Braiotta & Zhou 2006 audit committee alignment/ change Brown & Higgins 2001 earnings surprise management Butler et al. 2004 auditor opinions Carcello & Neal 2003 management discussion of financial distress Charitou et al. 2007 auditor opinions Chen et al. 2001 auditor opinions Chia et al. 2007 auditor size in different models Chin et al. 2006 earnings forecast Chung & Kallapur 2003 client importance Cohen et al. 2007 literature review Cohen et al. 2004 literature review Davidson & Neu 1993 earnings forecast error Davidson et al. 2006 auditor changes DeFond 1992 auditor changes DeFond & Jiambalvo 1991 accounting errors DeFond & Park 2001 abnormal accruals DeZoort et al. 2003 behavior experiment El Mir & Seboui 2006 market value Ettredge et al. 1988 stock return Fairfield et al. 2003 accrual growth Fields & Keys 2003 literature review Filatotchev et al. 2005 financial performance Francis 2006 literature review Francis et al. 2004 earnings attributes Gaver & Paterson 2001 loss adjustment Geiger et al. 2005 hiring of former auditors Givoly et al. 2007 reporting conservatism Glaum et al. 2004 firm characteristics Goodwin & Seow 2002 behavior experiment Gul et al. 2003 stock market reaction Audit Quality, Corporate Governance, and Earnings Management: A Meta-Analysis 63 Int. J. Audit. 14: 57–77 (2010)© 2009 Blackwell Publishing Ltd Table 2: Continued Study Different independent variable Different dependent variable No applicable data Gul et al. 2006 stock return Habib 2006 literature review Hartnett 2006 earnings forecast Haw et al. 2005 stock return Healy & Wahlen 1999 literature review Heninger 2001 auditor litigation Higgs & Skantz 2006 stock return Hodge 2003 earnings quality Hui & Fatt 2007 literature review Iturriaga & Hoffmann 2005 stock ownership Jenkins et al. 2006 interactive effect of event and auditor specialization main auditor specialization effect not reported Johl et al. 2007 auditor opinions Kanagaretnam et al. 2007 info asymmetry Karamanou & Vafeas 2005 earnings forecast Keasey & McGuinness 1991 earnings forecast Kim et al. 2003 auditor conservatism Knechel & Payne 2001 audit report lag Knechel & Vanstraelen 2007 auditor opinions Koh 2003 institutional ownership Krishnan 2003b stock market reaction Kwak 2002 literature review Lapointe-Antunes et al. 2006 voluntary disclosure Larcker & Richardson 2004 three subsample clusterwise regressions; no control variables Larcker et al. 2007 IVs are factor loading scores Lee & Mande 2003 private securities litigation reform act Lee et al. 2003 auditor choice Leng 2004 return on equity Leuz et al. 2003 investor protection Marnet 2007 literature review Matsumoto 2002 firm characteristics McNichols 2000 literature review Menon & Williams 1994 audit committee characteristics Mitra & Cready 2005 institutional ownership Nelson et al. 2003 survey Palmrose & Scholz 2004 restatement Peasnell et al. 2000 different accrual models Pergola 2005 literature review Petra 2007 stock market reaction Phillips et al. 2003 deferred tax expense Pincus & Rajgopal 2002 firm char. Rezaee et al. 2003 literature review Rowland 2002 literature review Ruddock et al. 2006 reporting conservatism Saleh & Ahmed 2005 debt renegotiation 64 J. W. Lin and M. I. Hwang Int. J. Audit. 14: 57–77 (2010)© 2009 Blackwell Publishing Ltd control mechanism. From an agency perspective, the ability of the board to function as an effective oversight of management in the areas of financial reporting rests upon its independence from management (Beasley, 1996). Therefore, it is assumed that effective governance and financial reporting quality increase with board independence (as measured by the proportion of outside or independent directors on the board). A slight majority of prior studies report a significantly negative relationship between earnings management and increased BOD independence (e.g., Beasley, 1996; Klein, 2002). However, Park & Shin (2004) and Peasnell et al. (2005) do not find a significant relationship. The meta-analytical results reported in Table 3 show that independence of the board has a significant negative relationship (at the 1% level) with occurrence of earnings management, based on either unweighted or weighted Stouffer test. Also, the fail-safe number greatly exceeds the critical number of studies (2,285 versus 100), hence strongly supporting the hypothesis that earnings management decreases as the board independence increases as suggested by Fama & Jensen (1983). BOD expertise While a more independent board may intend to restrain earnings management, only outside or independent directors on the board with proper background may be able to do so. A director with financial expertise may have greater familiarity with how earnings can be managed and take necessary measures to curb earnings management. Relatively few existing studies examine this issue. Table 2: Continued Study Different independent variable Different dependent variable No applicable data Sánchez-Ballesta & García-Meca 2005 audit opinions Sánchez-Ballesta & García-Meca 2007 stock ownership structure Schipper 1989 literature review Shaw 2003 corporate disclosure ratings Shen & Chih 2005 firm characteristics Srinidhi & Gul 2007 t or p-values not reported Srinivasan 2005 restatement Staubus 2005 literature review Summers & Sweeney 1998 insider trading Teoh & Wong 1993 stock market reaction Vafeas 2005 small earnings increase & unexpected earnings Van Caneghem 2004 earnings rounding-up behavior Van der Zahn & Tower 2006 auditor fees Wang 2006 founding family ownership Wells 2002 CEO changes Wild 1996 stock market reaction Wright et al. 2006 firm characteristics Xie 2001 stock market reaction Yee 2006 analytical study Yeo et al. 2002 management stock ownership Zhao & Millet-Reyes 2007 stock return Note: Studies are excluded if the independent variable is not defined as some measure of board or audit committee characteristics or audit quality, or the dependent variable is not based on abnormal accrual, financial statement restatement or fraud as the measure of earnings management (quality), with the variables actually used noted in the applicable cells. Studies are also excluded when the test statistics, p-value or similar data needed to compute the effect size for meta-analysis is not reported. Audit Quality, Corporate Governance, and Earnings Management: A Meta-Analysis 65 Int. J. Audit. 14: 57–77 (2010)© 2009 Blackwell Publishing Ltd One of the studies (Park & Shin, 2004) reports a significantly negative association between increased board financial expertise and earnings management, but another study (Xie et al., 2003) finds a negative but insignificant relationship. Our meta-analysis results suggest that the board financial expertise is negatively related to earnings management (significant at the 1% level) using either the unweighted or weighted Stouffer combined test. The fail-safe number exceeds the critical number of studies (38 versus 25), supporting the negative relationship. BOD stock ownership A clear theoretical prediction about the effect of stock ownership by directors on the effectiveness of the board in monitoring management does not exist. Gul et al. (2002: 30) argue that ‘managers of firms with low director ownership are expected to respond to accounting-based contracts by exploiting the latitude available in accounting procedures to either alleviate constraints or capitalize on available incentives suggesting a higher level of earnings management’. In other words, higher stock ownership by directors will reduce the occurrence of earnings management. However, a direct financial interest, such as stock ownership by outside directors, may weaken the independence of directors and their effectiveness in monitoring management decisions, including in the area of financial reporting. Prior studies provide mixed evidence on the effect on earnings management of directors’ stock ownership in the firm. For example, Gul et al. (2002) document a significantly negative association between directors’ stock ownership and earnings management. In contrast, Peasnell et al. (2005) report a positive, though not significant, association. Our meta-analysis suggests no significant relationship exists between stock ownership by directors and earnings management. Further research is probably warranted. BOD independent chair Another important characteristic of the board is whether the position of the Chief Executive Officer (CEO) is separate from the position of the chairperson of the board. One of the important roles played by the chairperson of the board is to run the board meetings and oversee the process of hiring, evaluating, firing and compensating the CEO. Jensen (1993) argues that it creates a conflict of interest for the CEO to serve as the board chair and perform the oversight function related to this Table 3: Effect of audit quality and corporate governance variables on earnings management Variable Stouffer test using unweighted Z Stouffer test using weighted Z Number of studies Fail- safe number Critical number for drawer Board of Directors (BOD) Independence -4.904*** -4.386*** 18 2285 100 BOD Expertise -3.655*** -3.511*** 3 38 25 BOD Stock Ownership 1.169 0.666 12 N/A N/A BOD Independent Chair 1.151 -0.271 10 N/A N/A Existence of Audit Committee (AC) -1.254 1.037 6 N/A N/A AC Independence -4.373*** -3.820*** 14 1043 80 AC Number of Meetings -3.788*** -2.487*** 10 218 60 AC Size -2.742*** -2.265** 7 85 45 AC Expertise -3.812*** -2.444*** 9 169 55 AC Stock Ownership 2.940*** 2.983*** 4 48 30 Auditor Tenure -3.264*** -3.095*** 7 166 45 Auditor Size -3.567*** -5.501*** 23 5892 125 Auditor Specialization -5.438*** -4.901*** 3 77 25 Auditor Independence Fee ratio 6.137*** 5.423*** 10 1076 60 Total fee 4.330*** 3.101*** 7 167 45 Audit fee 1.936** 0.232 4 N/A N/A Nonaudit fee 2.542*** 1.557 7 N/A N/A Note: ** = significant at the 5% level, *** = significant at the 1% level. 66 J. W. Lin and M. I. Hwang Int. J. Audit. 14: 57–77 (2010)© 2009 Blackwell Publishing Ltd [...]... et al., 2003: 71) In this meta-analysis, we review the relationships between earnings management and several attributes of audit quality commonly investigated in prior studies Int J Audit 14: 57–77 (2010) Audit Quality, Corporate Governance, and Earnings Management: A Meta-Analysis Auditor tenure Prior research suggests that auditor independence decreases as the length of auditor tenure increases (Beck... using unweighted Z Stouffer test using weighted Z Number of studies Fail-safe number Critical number for drawer -2 .848*** -3 .992*** -1 .892** -3 .684*** 6 12 42 710 40 70 0.18 1.706** 5 7 N/A N/A N/A N/A -3 .987*** -1 .968** -2 .329*** -2 .194** 9 5 153 39 55 35 -5 .065*** -0 .51 -5 .546*** -0 .414 9 14 911 N/A 55 N/A -3 .352*** -2 .764*** -1 .871** -2 .447*** 6 9 41 170 40 55 1.016 0.672 Note: ** = significant at the... Y., Lin, K-L & Zhou, J (2005), Audit quality and earnings management for Taiwan IPO firms’, Managerial Auditing Journal, Vol 20, No 1, pp 86–104 Chia, Y M., Lapsley, I & Lee, H-W (2007), ‘Choice of auditors and earnings management during the Asian financial crisis’, Managerial Auditing Journal, Vol 22, No 2, pp 177–96 Chin, C., Kleinman, G., Lee, P & Lin, M-F (2006), Corporate ownership structure and. .. accuracy and bias of mandatory earnings forecast: evidence from Taiwan’, Journal of International Accounting, Vol 5, No 2, pp 41–62 Choi, J., Jeon, K & Park, J (2004), ‘The role of audit committees in decreasing earnings statement: Int J Audit 14: 57–77 (2010) Audit Quality, Corporate Governance, and Earnings Management: A Meta-Analysis Korean evidence’, International Journal of Accounting, Auditing... the effect of auditor size as a deterrent to earnings management is more pronounced in the US than in other countries In fact, the effect is not Int J Audit 14: 57–77 (2010) Audit Quality, Corporate Governance, and Earnings Management: A Meta-Analysis significant when examined in other countries Also, the share ownership of the board of directors has no significant effect on earnings management, in... management , Auditing: A Journal of Practice & Theory, Vol 23, pp 13–35 Beekes, W., Pope, P & Young, S (2004), ‘The link between earnings and timliness, earnings © 2009 Blackwell Publishing Ltd J W Lin and M I Hwang conservatism and board composition: evidence from the UK’, Corporate Governance, Vol 12, No 1, pp 47–59 Benkel, M., Mather, P & Ramsay, A (2006), ‘The association between corporate governance and earnings. .. between audit firm size and audit quality’, Contemporary Accounting Research, Vol 9, No 2, pp 479–88 Davidson, R., Goodwin-Steward, J & Kent, P (2005), ‘Internal governance structures and earnings management , Accounting and Finance, Vol 45, pp 241–67 Davidson III, W N., Jiraporn, P & DaDalt, P (2006), ‘Causes and consequences of audit shopping: an analysis of auditor opinions, earnings management, and auditor... (2007), ‘Auditor fees and audit quality’, Managerial Auditing Journal, Vol 22, No 8, pp 761–86 Huang, H-W., Mishra, S & Raghunandan, K (2007), ‘Types of nonaudit fees and financial reporting quality’, Auditing: A Journal of Practice and Theory, Vol 26, No 1, pp 133–45 Hui, L T & Fatt, Q K (2007), ‘Strategic organizational conditions for risks reduction and earnings management: a combined strategy and auditing... Skantz, T R (2006), Audit and nonaudit fees and the market’s reaction to earnings announcements’, Auditing: A Journal of Practice & Theory, Vol 25, No 1, 1–26 Hodge, F D (2003), ‘Investors’ perceptions of earnings quality, auditor independence, and the usefulness of audited financial information’, Accounting Horizons, Vol 17, No 1, pp 37–48 © 2009 Blackwell Publishing Ltd J W Lin and M I Hwang Hoitash,... larger audit committee represents greater resources and talents to rely on in overseeing the financial reporting Int J Audit 14: 57–77 (2010) J W Lin and M I Hwang 68 process Empirical studies provide mixed evidence on the impact of audit committee size on earnings management Xie et al (2003) find no significant association between the number of directors on the audit committee and earnings management . Audit Quality, Corporate Governance, and Earnings Management: A Meta-Analysis ija_403 57 78 Jerry W. Lin 1 and Mark I. Hwang 2 1 University of Minnesota Duluth 2 Central Michigan University Earnings. ratio and total fee, is also a deterrent to earnings management. Key words: Audit committee, audit quality, auditor choice, corporate governance, earnings management, fraud, independence, meta-analysis SUMMARY Earnings. Mayhew, B. W. (2003), ‘Do nonaudit services compromise auditor Audit Quality, Corporate Governance, and Earnings Management: A Meta-Analysis 71 Int. J. Audit. 14: 57–77 (2010) © 2009 Blackwell Publishing